Printer Friendly

The Australian soil texture boomerang: a comparison of the Australian and USDA/FAO soil particle-size classification systems.

Introduction

Various pedotransfer functions for predicting soil physical and chemical properties have been developed. Cresswell and McKenzie (2000) developed a computer program for predicting soil hydraulic properties of Australian soil. However, the different classification of particle-size fractions used in Australia compared with other countries presents a problem for the immediate adoption of the exotic pedotransfer functions. Nemes et al. (1999a) encountered this problem when evaluating different textural classes used in various European countries (Fig. 1).

[FIGURE 1 OMITTED]

There are 2 major textural classifications used in the world--the International and the USDA/FAO systems. The particle-size limits are:
 International USDA/FAO

clay <2 [micro]m <2 [micro]m
silt 2-20 [micro]m 2-50 [micro]m
sand 20-2000 [micro]m 50-2000 [micro]m


The first classification, the International system, was first proposed by Atterberg (1905), and was based on his studies in southern Sweden. Atterberg (1908) chose 20 [micro]m for the upper limit of silt fraction because particles smaller than that size were not visible to the naked eye, the suspension could be coagulated by salts, capillary rise within 24 h was most rapid in this fraction, and the pores between compacted particles were so small as to prevent the entry of root hairs. Commission One of the International Society of Soil Science (ISSS) recommended its use at the first International Congress of Soil Science in Washington in 1927 (International Society of Soil Science 1929). Australia adopted this system, and according to Marshall (1947) its equal logarithmic intervals are an attractive feature worth maintaining. The USDA adopted its own system in 1938 (Knight 1937), and the FAO used the USDA system in the FAO-UNESCO world soil map (FAO-UNESCO 1974) and recommended its use (FAO 1990).

Soil texture represents the relative composition of sand, silt, and clay in soil. The particle-size distribution is usually represented in a texture diagram, relating the percentages of sand, silt, and clay to a texture class. According to Marshall (1947) the first texture diagram relating the particle-size distribution to textural classes was drawn in 1911 by Whitney (1911) where the clay and silt percentages were represented on a right-angled triangle. Davis and Bennett (1927) replaced it by a more difficult-to-read ternary diagram which shows percentages of clay, silt, and sand. In 1945, the USDA revised its texture triangle to accommodate changes in analytical procedure and particle-size limits.

Prescott et al. (1934) devised a texture diagram for the International system based on mechanical analysis and field descriptions made by the CSIR (Council for Scientific and Industrial Research) Division of Soils at that time. This was later revised by Marshall (1947) and has been used in Australia until the present (McDonald et al. 1990). Since there are differences in the particle-size limits, the textural classes between the USDA and Australian system are also not synonymous.

Toogood (1958) presented a `simplified' Canadian texture diagram (which is similar to the USDA) in which percentages of sand and clay were plotted on a right triangle. McBratney suggested this type of diagram to be used in England and Wales (Mullins 1991).

Particle-size conversion

Generally log-linear interpolation on the cumulative particle-size distribution is used to estimate missing particle-size classes for a given classification (Tietje and Hennings 1996). This method assumes that particle-size is lognormally distributed (Shirazi and Boersma 1984). Shirazi et al. (1988) established a conversion table between the USDA texture classifications and the International system. Their conversion assumed a lognormal particle-size distribution assumption within each size fraction (sand, silt, and clay). Buchan (1989) also proposed conversion of the USDA texture to the International system based on a lognormal particle-size distribution model. Rousseva (1997) defined several closed-form exponential and power functions and investigated the suitability of these models in describing cumulative particle-size distributions. Suitability of the models appeared to be influenced by textural type rather than number of measured points and size ranges. Nemes et al. (1999a) evaluated 4 different interpolation methods (log-linear interpolation, fitting a Gompertz curve, spline interpolation, and similarity method) in order to achieve compatibility of particle-size distributions within the European soil hydraulic database HYPRES. They introduced the similarity procedure, which uses an external reference data set that contains several soil materials with 7 or 8 measured particle-size fractions. The procedure involves searching for soil samples in the external reference data set that match the particle-size distribution of the soil to be interpolated. This procedure was found to give the most accurate interpolations.

Because in routine soil survey, particle-size is measured only on 3 or 4 fractions (clay, silt, fine sand, coarse sand), it is not appropriate to fit a model that has 3 or 4 parameters. An empirical regression seems more suitable, e.g. Marshall (1947) approximated the conversion as:

[P.sub.2-20] = 0.5 [P.sub.2-50] [+ or -] 0.13 [P.sub.2-50]

where [P.sub.2-20] and [P.sub.2-50] are the proportional masses of particles with diameter 2-20 [micro]m and 2-50 [micro]m, respectively. Minasny et al. (1999) converted the 2-20 [micro]m fraction into a 2-50 [micro]m fraction based on the Australian data sets that provide measurements for both fractions. The model calculated the increment of particles when converting 2-20 [micro]m to 2-50 [micro]m fractions ([P.sub.20-50]):

(1) [P.sub.20-50] (dag/kg) = 48.4593 - 0.2225 [P.sub.20-2000] - 0.0029 [([P.sub.20-2000]).sup.2] - 0.6952 [P.sub.<2] + 0.0018 [([P.sub.<2]).sup.2] ([R.sup.2] = 0.76, n = 290)

In order to achieve better prediction, a larger data set has been compiled. The data set is derived from soil databases which contain measurement of both fractions, i.e. the Australian databases (Minasny et al. 1999), UNSODA (Nemes et al. 1999b), GRIZZLY (Haverkamp et al. 1997), and the USDA-NRCS Soil Survey Laboratory data (USDA-NRCS 1997). All data were combined and randomly split into two: a prediction set (n = 1210) and a validation set (n = 400). Particle-size conversion was modelled using a multiple linear regression. The model is:

(2) [P.sub.2-50] = -18.3914+ 2.0971 ([P.sub.2-20]) + 0.6726 ([P.sub.20-2000) - 0.0142 [([P.sub.2-20]).sup.2] - 0.0049 [([P.sub.20-2000]).sup.2] ([R.sup.2] = 0.823)

If [P.sub.2-50] < 0 then [P.sub.2-50] = 0.8289 ([P.sub.2-20]) + 0.0198 ([P.sub.20-2000])

Conversely, if we had measurements based on the USDA/FAO classification, a model was derived to predict [P.sub.2-20]:

(3) [P.sub.2-20] = -0.4070 - 0.1271 ([P.sub.<2]) + 0.5527 ([P.sub.2-50]) + 0.0017[([P.sub.<2]).sup.2] - 0.0019 [([P.sub.2-50]).sup.2] + 0.0059 ([P.sub.<2]) [([P.sub.2-50]) ([R.sup.2] = 0.818) If [P.sub.2-20] < 0 then [P.sub.2-20] = 0.1154 ([P.sub.<2]) + 0.2212 ([P.sub.2-50])

Table 1 presents the mean error (ME) and root mean squared error (RMSE) between the predictions and measured values. A log-linear interpolation was also tried by way of comparison:

(4) C[P.sub.i] = C[P.sub.i-1] + C[P.sub.i+1] - C[P.sub.i-1]/log(P[S.sub.i+1])- log(P[S.sub.i-1])[log(P[S.sub.i])-log(P[S.sub.i-1])]

where CP is the cumulative amount of panicles mass on PS particle-size limit, and i, i-1, i+1 represent the point to interpolate, and the preceding and succeeding neighbour limits (Nemes et al. 1999a). The results showed that the regression is clearly better than the log-linear interpolation model. Predicting [P.sub.2-20] is more accurate ([+ or -] 5%) than predicting [P.sub.2-50] ([+ or -] 8%).

Textural-class conversion

The small range of silt in the International system and the non-existence of the silt texture in the Australian texture triangle posed a question how the textural classes of the USDA/ FAO system correspond to those of Australia. Marshall (1947) indicated that it was impossible to make direct comparison between texture on the two systems. Marshall (1947) presented a `rough conversion' of the USDA textural classes into the International system coordinates. Since it is based on a linear transformation ([P.sub.2-20] = 0.5[P.sub.2-50]) the texture classes were shown to occupy exactly one-half of the International system's triangle. Using Eqn 3, the vertices from each of the textural classes from the USDA/FAO soil texture triangle (Soil Survey Staff 1975) were convened to the International system.

Figure 2a shows that the USDA/FAO textural class is approximately a `boomerang' shape in the International system. Table 2 shows the centroid of the polygon for each textural class and the percentage of the area occupied on the triangle and the `boomerang'. The total area is about 60% of the triangle, 10% more than predicted by Marshall (1947). Likewise, Fig. 2b shows the Australian texture classes plotted in the USDA/FAO system. Figure 3 shows the USDA texture classes plotted together with the Australian system. There are overlaps between the 2 textural classes, especially for the sandy clay and sandy clay loam textures.

[FIGURES 2-3 OMITTED]

In addition to the 1610 soil samples mentioned above that contain measurements on both systems, 820 soil samples obtained later from USDA-NRCS Soil Survey Laboratory data were plotted on the two texture triangles (Fig. 4). The data are more evenly distributed in the USDA/FAO system, while in the International system most of the data are within the `boomerang'. Very few data fell outside the boomerang, in this case only 5%. This `blank' region represents soil that has [P.sub.2-20] > 40%. It is uncommon to find a soil which has such a large mass in the 2-20 [micro]m range. The non-parametric quantile density plots show that the data are distributed in the middle of the USDA/FAO texture triangle, while in the International triangle the data are only within the `boomerang'.

[FIGURE 4 OMITTED]

A textural class contingency table of the USDA/FAO and Australian systems (Table 3) shows that the USDA classes overlap to a fair degree with the Australian classes, which confirms the representation in Fig. 3. Based on Table 3, sandy loam (USDA/FAO) can translate into sand, loamy sand, sandy loam, loam, silt loam, and sandy clay loam in the Australian system. Silt loam (USDA) can translate into 8 other classes in the Australian system.

Some overlaps should not have occurred according to Fig. 3 (e.g. silt loam to loam). This is because the conversion is based on an approximate regression and in reality the particle-size distributions can be multi-modal (Walker and Chittleborough 1986) and the data are also bound to contain a degree of measurement error.

Conclusions

We note (1) no very strong argument for the use of a 2-20 [micro]m, as opposed to a 2-50 [micro]m, silt fraction has been put forward; (2) the very uneven distribution of observations in the International/Australian system and the vast empty area in the lower left of the texture triangle; and (3) the much more evenly and widely populated distribution of observed soil textures in the triangle of the USDA/FAO system.

We therefore suggest on scientific and educational grounds that it would seem wise for most countries, including Australia, to consider adopting the particle-size limits and texture classes of the USDA/FAO system. With computing power, databases, and good algorithms this is much less important than in the past--translations from one system to another may be made. Nevertheless, the process of gathering soil data is by no means over and the generation of new data with appropriate particle-size limits is important.

We note that the 2-50 [micro]m particle size range is usually more useful than 2-20 [micro]m for estimating water retention in pedotransfer functions. We also note that currently there are particle-size analysers used in Australian laboratories that can provide 10-15 size classes, which allows for a continuous particle-size distribution. However, in order to use this information effectively, we need to parameterise the distribution; a parsimonious yet flexible function that can accommodate multi-modal size distributions is required.

Finally, if we call the limit between `clay' and `silt' [phi] and that between `silt' and `sand' [kappa], e.g. in the USDA/FAO System [phi] is 2 [micro]m and [kappa] is 50 [micro]m, it might be instructive to optimise the values of [phi] and [kappa] to produce a more even and widespread distribution of points in the triangle. Less radically, since most countries, with the possible sole exception of Slovakia, have adopted a [phi] of 2 [micro]m, this optimisation could be performed for [kappa] alone.
Table 1. Statistics of the error in particle-size conversion

 Predicted Equa-
Method particle- tion Prediction set Validation set
 size no. (n = 1210) (n = 400)

 ME RMSE ME RMSE

Regression [P.sub.2-20] (3) -0.01 4.63 0.26 5.38
Log-linear [P.sub.2-20] (4) -6.06 8.73 6.10 9.08
Regression [P.sub.2-50] (2) 0.00 7.48 0.39 8.01
Log-linear [P.sub.2-50] (4) 1.60 11.96 -1.90 13.10
Table 2. Centroids of the USDA textural classes in the
International system

 Centroid

Texture [P.sub. [P.sub. % Area % Area
class 20-2000] 2-20] [P.sub.<2] triangle boomerang

Sand 94.1 2.6 3.3 0.8 1.3
Loamy sand 87.8 6.3 5.9 1.5 2.5
Sandy loam 77.1 12.3 10.7 5.7 9.5
Loam 59.9 21.4 18.7 3.9 6.5
Silt loam 53.8 32.1 14.0 6.5 10.9
Silt 57.4 37.1 5.5 0.9 1.5
Sandy clay
 loam 65.8 7.0 27.1 4.2 7.0
Clay loam 46.9 19.3 33.8 4.1 7.0
Silty clay
 loam 31.5 34.4 34.0 3.7 6.1
Sandy clay 54.7 3.6 41.7 2.2 3.6
Silty clay 20.1 33.3 46.7 3.5 5.9
Clay 24.0 12.8 63.3 22.8 38.2

Total 59.6
Table 3. Contingency table showing the distribution of the
USDA texture classes with the International system

Values are percentage of the number of data

Australia USDA/FAO

 S LS SL L ZL Z SCL

S 12.2 2.2 0.2 0.0
LS 0.5 6.6 14.8 1.8 0.2
SL 0.0 0.6 4.6 0.2 0.1 0.0
L 4.4 5.9 2.4 0.5
ZL 1.2 2.5 7.7 0.2
[Z]
SCL 0.1 0.4
CL 1.1 0.2 0.9
ZCL 0.2 0.3
SC 0.1
ZC 0.1 0.5
C 0.2 0.1

Sum 12.8 9.4 25.3 12.1 11.4 0.2 1.9

Australia USDA/FAO Sum

 CL ZCL SC ZC C

S 14.7
LS 23.9
SL 5.6
L 13.2
ZL 11.5
[Z]
SCL 0.5
CL 2.8 0.2 5.1
ZCL 1.4 4.7 6.5
SC 0.1
ZC 0.1 0.1 3.3 2.0 6.2
C 1.4 0.2 0.3 0.8 9.7 12.6

Sum 5.7 5.2 0.3 4.1 11.7


References

Atterberg A (1905) Die rationalle Klassifikation der Sande und Kiese. Chemiker Zeitung 29, 195-198.

Atterberg A (1908) Studien auf dem Gebiete der Bodenkunde. Landwirtschaftliche Versuchs-Stationen 69, 93-143.

Buchan GD (1989) Applicability of the simple lognormal model to particle-size distribution in soils. Soil Science 147, 155-161.

Cresswell HP, McKenzie NJ (2000) A toolkit for hydraulic properties of Australian soils. Australian Collaborative Land Evaluation Program Newsletter No. 8, 15-10.

Davis ROE, Bennett HH (1927) Grouping of soils on the basis of mechanical analysis. United States Department of Agriculture Departmental Circulation No. 419.

FAO-UNESCO (1974) `FAO-UNESCO Soil Map of the World: 1:5 000 000.' Prepared by the Food and Agriculture Organization of the United Nations (UNESCO: Paris)

FAO (1990) `Guidelines for soil description.' 3rd edn. (FAO & ISRIC: Rome)

Haverkamp R, Zammit C, Bouraoui F, Rajkai K, Arrue JL, Heckman N (1997) `GRIZZLY, Grenoble Soil Catalogue. Soil survey of field data and description of particle size, soil water retention and hydraulic conductivity functions.' (Laboratoire d'Etude des Transfers en Hydrologie et Environnement, LTHE: Grenoble Cedex 09, France)

International Society of Soil Science (1929) Minutes of the first commission meetings, International Congress of Soil Science. Transactions of the First Commission of the International Society of Soil Science 4, 215-220.

Knight HG (1937) New size limits for silt and clay. Soil Science Society of America Proceedings 2, 592.

Marshall TJ (1947) Mechanical composition of soil in relation to field descriptions of texture. Council for Scientific and Industrial Research, Bulletin No. 224, Melbourne.

McDonald RC, Isbell RF, Speight JG, Walker J, Hopkins MS (1990) `Australian soil and land survey field handbook.' 2nd edn (Inkata Press: Melbourne)

Minasny B, McBratney AB, Bristow KL (1999) Comparison of different approaches to the development of pedotransfer functions for water-retention curves. Geoderma 93, 225-253.

Mullins CE (1991) Physical properties of soils in urban areas. In `Soils in the urban environment'. (Eds P Bullock, PJ Gregory) pp. 87-118. (Blackwell Scientific Publications: Oxford)

Nemes A, Wosten JHM, Lilly A, Oude Voshaar JH (1999a) Evaluation of different procedures to interpolate particle-size distributions to achieve compatibility within soil databases. Geoderma 90, 187-202.

Nemes A, Schaap MG, Leij FJ (1999b) `The UNSODA unsaturated soil hydraulic database Version 2.0.' (US Salinity Laboratory: Riverside, CA)

Prescott JA, Taylor JK, Marshall TJ (1934) The relationship between the mechanical composition of the soil and the estimate of texture in the field. Transactions of the First Commision of the International Society of Soil Science 1, 143-153.

Rousseva SS (1997) Data transformations between soil texture schemes. European Journal of Soil Science 48, 749-758.

Shirazi MA, Boersma L (1984) A unifying quantitative analysis of soil texture. Soil Science Society of America Journal 48, 142-147.

Shirazi MA, Boersma L, Hart JW (1988) A unifying quantitative analysis of soil texture: improvement of precision and extension of scale. Soil Science Society of America Journal 52, 181-190.

Soil Survey Staff (1975) `Soil taxonomy: A basic system of soil classification for making and interpreting soil surveys'. (United States Department of Agriculture, Soil Conservation Service)

Tietje O, Hennings V (1996) Accuracy of the saturated hydraulic conductivity prediction by pedo-transfer functions compared to the variability within FAO textural classes. Geoderma 69, 71-84.

Toogood JA (1958) A simplified textural classification diagram. Canadian Journal of Soil Science 38, 54-55.

USDA-NRCS (1997) National Soil Characterization Database. USDA-NRCS Soil Survey Division. http:// www.statlab.iastate.edu/soils/ssl/natch_data.html.

Walker PH, Chittleborough DJ (1986) Development of particle-size distributions in some Alfisols of southeastern Australia. Soil Science Society of America Journal 50, 394-400.

Whitney M (1911) The use of soils east of the Great Plains region. United States Department of Agriculture Bureau of Soils Bulletin No. 78.

Manuscript received 16 August 2000, accepted 5 April 2001

Budiman Minasny and Alex. B. McBratney

Department of Agricultural Chemistry and Soil Science, The University of Sydney, Ross St. Building A03, NSW 2006, Australia.
COPYRIGHT 2001 CSIRO Publishing
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2001 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Minasny, Budiman; McBratney, Alex. B.
Publication:Australian Journal of Soil Research
Article Type:Statistical Data Included
Geographic Code:8AUST
Date:Nov 1, 2001
Words:3188
Previous Article:Comparing simple methods for measuring phosphate sorption by soils.
Next Article:The burrowing characteristics of three common earthworm species.
Topics:


Related Articles
Soil erodibility ([K.sub.m]) values for some Australian soils.
Estimation of erosion model erodibility parameters from media properties.
Soil stripping and replacement for the rehabilitation of bauxite-mined land at Weipa. II. Soil organic matter dynamics in mine soil chronosequences.
Soil stripping and replacement for the rehabilitation of bauxite-mined land at Weipa. III. Simulated long-term soil organic matter development.
The red clay mantle in the Wagga Wagga region, New South Wales: evaluation of an aeolian dust deposit (Yarabee Parna) using methods of soil landscape...
The influence of aeolian dust deposits on alpine soils in south-eastern Australia.
Field prediction of sodicity in dryland agriculture in Central Queensland, Australia.
Dispersibility of glacial loess in particle size analysis, USA.
Particle-size distribution of soil and the perception of texture. (Short Communication).
E[C.sub.1:5]/exchangeable Na, a sodicity index for cotton farming systems in irrigated and rainfed Vertosols. (Short Communication).

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters